如何阻止ansible从跳到第二个任务直到成功执行第一个任务？

如何解决如何阻止ansible从跳到第二个任务直到成功执行第一个任务？

我的剧本看起来像这样：

--- 
  - hosts: localhost
    gather_facts: false
  - name: execute task 1
    shell: nohup sh test001.sh >> nohup.out 2>&1 &
    async: 120
    poll: 10
  - name: read generated log for task 1
    shell: cat test001result*
    register: execution_status
  - name: If task 1 failed
    shell: nohup sh test002.sh >> nohup.out 2>&1 &
    when: execution_status.stdout.find('success') == -1
  - name: If task 1 passed
    shell: nohup sh test003.sh >> nohup.out 2>&1 &
    when: execution_status.stdout.find('success') != -1

现在任务1需要60秒才能完成，即至少在60秒后生成test001result。尽管添加了异步120秒，但无论task1成功如何，ansible执行都会继续，并且仍执行test002.sh。这是因为尚未生成日志（显示0字节）。

如何纠正此问题？

解决方法

给出脚本

shell> cat test001.sh
echo $(date) test_01 started
sleep 3
echo 'success' > test001result
echo $(date) test_01 finished
exit 0

shell> cat test002.sh
echo $(date) test_02 started
sleep 3
echo $(date) test_02 finished
exit 0

shell> cat test003.sh
echo $(date) test_03 started
sleep 3
echo $(date) test_03 finished
exit 0

在剧本中，不要分离过程

  nohup sh test001.sh >> nohup.out 2>&1 &

只需运行

  sh test001.sh >> nohup.out 2>&1

例如剧本

shell> cat playbook.yml
- hosts: localhost
  gather_facts: false
  tasks:
  - name: execute task 1
    shell: sh test001.sh >> nohup.out 2>&1
    async: 20
    poll: 5
    ignore_errors: true
  - name: read generated log for task 1
    shell: cat test001result
    register: execution_status
    ignore_errors: true
  - debug:
      var: execution_status.stdout
  - name: If task 1 failed
    shell: nohup sh test002.sh >> nohup.out 2>&1 &
    when: execution_status.stdout is not search('success')
  - name: If task 1 passed
    shell: nohup sh test003.sh >> nohup.out 2>&1 &
    when: execution_status.stdout is search('success')

给予

shell> ansible-playbook playbook.yml

PLAY [localhost] ****

TASK [execute task 1] ****
changed: [localhost]

TASK [read generated log for task 1] ****
changed: [localhost]

TASK [debug] ****
ok: [localhost] => 
  execution_status.stdout: success

TASK [If task 1 failed] ****
skipping: [localhost]

TASK [If task 1 passed] ****
changed: [localhost]

PLAY RECAP ****
localhost: ok=4 changed=3 unreachable=0 failed=0 skipped=1 rescued=0 ignored=0

shell> cat nohup.out 
Tue 25 Aug 2020 09:50:35 PM CEST test_01 started
Tue 25 Aug 2020 09:50:38 PM CEST test_01 finished
Tue 25 Aug 2020 09:50:41 PM CEST test_03 started
Tue 25 Aug 2020 09:50:44 PM CEST test_03 finished

如果test002.sh不成功，则应以同样的方式运行test001.sh。

---

指令async和poll用于运行可能不会在合理时间内结束的进程的目的。让我们测试一下这种情况，并在test001.sh中将睡眠增加到30秒。在这种情况下，前两个任务都将失败。任务execute task 1将因async超时而失败，而任务read generated log for task 1将因缺少文件test001result而失败。我们必须为这两个任务设置ignore_errors: true。现在，剧本给出了

shell> ansible-playbook playbook.yml

PLAY [localhost] ****

TASK [execute task 1] ****
fatal: [localhost]: FAILED! => changed=false 
  msg: async task did not complete within the requested time - 20s
...ignoring

TASK [read generated log for task 1] ****
fatal: [localhost]: FAILED! => changed=true 
  cmd: cat test001result
  delta: '0:00:00.003367'
  end: '2020-08-25 22:11:56.422448'
  msg: non-zero return code
  rc: 1
  start: '2020-08-25 22:11:56.419081'
  stderr: 'cat: test001result: No such file or directory'
  stderr_lines: <omitted>
  stdout: ''
  stdout_lines: <omitted>
...ignoring

TASK [debug] ****
ok: [localhost] => 
  execution_status.stdout: ''

TASK [If task 1 failed] ****
changed: [localhost]

TASK [If task 1 passed] ****
skipping: [localhost]

PLAY RECAP ****
localhost: ok=4 changed=2 unreachable=0 failed=0 skipped=1 rescued=0 ignored=2

shell> cat nohup.out 
Tue 25 Aug 2020 10:11:35 PM CEST test_01 started
Tue 25 Aug 2020 10:11:56 PM CEST test_02 started
Tue 25 Aug 2020 10:11:59 PM CEST test_02 finished